Talking to my Brain agent.

Hey 👋,

I am a heavy user of AI. And I have been playing around with a personal assistant solution for a while now. Basically I have my own repository with some instructions that accumulate knowledge the more I talk to it. I gave it a personality and it’s called Brain.

This was created after playing around with OpenClaw in the first days of its release. I didn’t want to have something I didn’t trust running on my computer, and I thought that just a folder with some instructions would be good enough. And it did. Anyway, this is not about Brain itself (even though thinking about it now I feel like sharing more about it!).

The thing, last night I decided I would start walking to the gym (instead of driving). It’s just a 20 minute walk anyway but I always procrastinate until the last minute to leave home so it’s always more convenient to drive so I am not late (or too late).

So, to be consistent with myself, I spent 15 minutes (which I didn’t have) getting conversation mode working with my agent via ChatGPT. It’s something I didn’t try before even though I knew all the pieces were there.

Sorry about this ☝️. Do you like my orange trainers? I love them.

The setup was like this:

• I have Codex (CLI) running at home. With an active session in my /brain repository. I keep one of those always open, so this is nothing new.
• I paired my ChatGPT app (on the phone) with the computer. I did this with codex remote-control start and codex remote-control pair (official docs for CLI here; with the full Codex app it’s even easier).
• On the phone then: ChatGPT > Codex > Kinton.local (my laptop’s hostname) and searched for the active session in the /brain folder. I clicked the conversation icon and started talking to it.

🧠 THOUGHT #1: What a chaos are becoming the AI ecosystems. There are a hundred of slightly different pieces you can combine into what look like pretty similar resulting solutions. This is pretty similar to what I would use Dots for, for example.

It was odd at first but, after the first few phrases, conversation just flowed naturally. It’s interesting interrupting and getting interrupted by an AI.

I was tempted to say that “it was a productive walk”. But I hated that whole phrase. So I’d better say “it was useful and fun”. Please, let’s stop obsessing about being productive every minute of our lives. Life is something more. If you haven’t found it yet, that’s ok too. Looking for it might be it.

🧠 THOUGHT #2: I go back and forth between Claude (Anthropic) and Codex (ChatGPT). I don’t know if Anthropic has an equivalent to “conversation mode” but I bet you can get a similar solution working there.

🧠 THOUGHT #3: This is an unusual style of post from me but I really enjoyed writing this! I even drew a sketch with an online Paint.

As always, if you liked this post leave a like below. Everytime I get a like a like a notification popsup for me and I get to smile. It makes my day :D